Zed
@zed.dev
3 months ago Agent message queuing is now the default. Queue up prompts while the agent is working, edit queued messages before they're sent, and interrupt the queue at any time.
💬 2
♻️ 0
❤️ 7
Agent message queuing is now the default. Queue up prompts while the agent is working, edit queued messages before they're sent, and interrupt the queue at any time.
🚀 We just shipped v0.221!
You can now install agents through the new ACP Registry—a unified hub for discovering and distributing ACP agents.
Check out our latest blog post for the details:
zed.dev/blog/acp-reg... Clark Wimberly had one of the best takeaways from our Community Showcase: QA is becoming the superpower in the agent era.
Humans still have to catch what models miss.
📺 Full video: www.youtube.com/watch?v=oKpF... GLM-4.7-Flash just dropped.
30B MoE running ~43 tokens/sec on an Apple M5.
Local LLMs are getting very real.
We shared a full walkthrough for building a local AI SaaS 👇
pris.ly/y/typefully... 2025 was a lot. Especially on the security front. 🔐
@philippeserhal.com tracked the biggest web framework themes (hello, CVEs) and what they mean going into 2026.
www.netlify.com/blog/web-fra... Did you know? In Vitest, a test without a callback function is automatically marked as `todo` and skipped. This is great for scaffolding out your test suite before you (or your LLM) write the actual tests, maybe even *before* the actual functionality is implemented 🙌🏻
Check out the docs!
pris.ly/docs/select You're probably fetching more data than you need.
Prisma ORM includes all fields by default.
Switching to select only what you need, can lead to major improvements in high traffic applications. i've been too nice about not linking the post, but in light of the last few days here ya go the dumbest bullshit imaginable nabeelqu.co/reflections-... running a small experiment here, but Atuin is now correctly reading the prompt/command/command output colour coded in debug mode let's see what we can do when we're no longer limited by shell support
Thank you so much to @planetscale.com for sponsoring Atuin on GitHub Sponsors!
Very much appreciate your support <3 Or, if you want to avoid any “oopsy” moments, our safety guide has you covered 👇
pris.ly/d/ai-safety wait...what ? Learn more in our docs:
planetscale.com/docs/postgre... Need Postgres connection pooling? PlanetScale has three PgBouncer options: - Local: co-located with the primary Postgres instance - Dedicated primary: runs on separate infrastructure, good for pooling + HA - Dedicated replica: pooling for your replica queries
We're exploring how to bring Tuist's binary caching to external build systems like Gradle, Cargo, and CMake.
If you're using Kotlin Multiplatform, Rust, or C++ in your iOS projects, this could save you a lot of rebuild time.
Would love your input on the RFC:
https://community.tuist.... Quick takeaway from @maddiedreese.bsky.social: use AI to move faster, not to check out.
Full Community Showcase on YouTube: www.youtube.com/watch?v=oKpF... If you hit plugin compatibility questions, reply here. We’re listening.
Gatsby is not “dead.” It shipped. 🎉
Gatsby v5.16.0 is out with official support for React 19 + Node.js 24.
Not a breaking change. You can upgrade Gatsby while staying on React 18, then move to React 19 when you’re ready.
Release notes:
www.gatsbyjs.com/docs/referen... Slow queries aren’t always obvious.
Prisma Optimize surfaces real query latency, variance, and hidden costs like cold connections. It runs in dev so you can catch issues early before they hit prod.
See what's slowing things down, and why.
pris.ly/docs/optimize 🔍 @rolldown.rs can now automatically discover your tsconfig, similar to how @vite.dev does it, respecting `references`.
The feature was available beforehand but is now enabled by default!
Don't want auto discovery? Then provide a path to another config
More in the docs
rolldown.rs/reference/In... We've seen our users struggle with setting up Prisma ORM and @supabase.com!
So we made a video about it, check it out👇
pris.ly/y/prisma-su... i came up with a new way to build software! you write the code yourself with your fingers it's claude code, in your own brain it's as local-first as it gets, and sustainable: just slop bowls & 1 gallon of water lmk if you'd like a help installing in your brain! it's called "learning to code"